Monitoring services often evaluate data from a variety of computer systems and hardware components of these computer systems to identify anomalous signals that may serve to indicate an issue with these computer systems or hardware components. However, identifying anomalous signals from a large amount of data obtained from these computer systems and hardware components presents various challenges. For instance, for modern monitoring services, it may be difficult to determine which data streams are important, how they relate to each other, and determine what in the data streams indicate an anomaly. Additionally, in order to reduce the impact of these anomalies, the monitoring services need to be able to identify anomalies in an expeditious manner.